AI-BLADE toolbox: AI-powered BLADdEr multiparametric MRI analysis for clinical application

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Growing need for user-friendly, bladder-specific AI tools for quantitative imaging biomarkers (QIBs) from multiparametric MRI (mpMRI).
• Current tools lack reliability for clinical applications in bladder cancer (BCa).

Purpose of the Study:

• To develop and validate AI-BLADE (AI-powered BLADdEr multiparametric MRI Analysis for Clinical Application), a novel toolbox for BCa mpMRI analysis.
• To extract reliable AI-QIBs for enhanced clinical decision-making.

Main Methods:

• AI-BLADE integrates Deep Feature Analysis (MRI-DFA) and Data-Driven Model-Based Analysis (MRI-MBA) toolkits.
• DFA classified BCa histology subtypes (n=104) using T2-weighted images.
• MBA derived mpMRI QIBs (ADC, Ktrans) from 34 BCa patients.

Main Results:

• The VGG19 model with a decision tree classifier achieved an AUC of 0.79 for BCa histology classification.
• Mean ADC values were 1.22 × 10⁻³ mm²/s and mean Ktrans values were 0.27 min⁻¹.
• AI-BLADE demonstrated strong performance in classifying BCa subtypes and deriving key physiological metrics.

Conclusions:

• AI-BLADE (v1.0) is a flexible, user-friendly software for mpMRI analysis in BCa oncology.
• The toolbox shows potential to enhance diagnostic accuracy and improve patient outcomes.
• This novel AI toolbox facilitates AI-QIB-based clinical decision-making for BCa patients.